Key takeaways
- An ontology defines classes, properties and relationships, plus rules about them.
- A taxonomy is the simpler cousin: a hierarchy of categories without the relationships and rules.
- OWL is the W3C standard language for writing ontologies a program can check.
- A knowledge graph holds the facts; the ontology says what kinds of facts are allowed.
- Shared definitions of account, opportunity and customer keep reports and AI summaries counting the same things.
What an ontology contains
An ontology contains the classes of things in a domain, the properties they have, the relationships allowed between them and the rules that must hold. Together these make a vocabulary precise enough for a program to apply, not only for people to read.
- Classes: the kinds of thing, such as account, contact, opportunity and product.
- Properties: what each has, such as an account's industry or a deal's stage.
- Relationships: how things connect, such as a contact works at an account.
- Rules: what must be true, such as a customer is an account with at least one closed-won opportunity.
Ontology vs taxonomy
An ontology goes further than a taxonomy: a taxonomy arranges categories in a hierarchy, while an ontology also defines relationships across the hierarchy and rules about them. An industry classification, with software under technology and payroll software under software, is a taxonomy.
Turn it into an ontology and you add statements such as 'a company operates in one or more industries' and 'a company in a regulated industry needs a compliance review before contract'. Most ontologies contain a taxonomy-like backbone; most taxonomies are not ontologies.
Ontology vs knowledge graph
An ontology is the schema and a knowledge graph is the data: the ontology says an account can have a parent company, and the graph records that this account's parent is that company. A graph built without an ontology still works, but different people tend to describe the same relationship in different ways.
Pairing the two is what lets software check a graph for contradictions, such as an account recorded as both a customer and a never-contacted prospect, and infer facts the data implies but does not state.
OWL ontology: the standard format
An OWL ontology is one written in the Web Ontology Language, the W3C standard for defining classes, their members and the relationships allowed between them in a form with exact logical meaning. Because the language has defined logical meaning, programs called reasoners can check an OWL ontology for consistency and derive facts it implies.
Most business teams will never write OWL. It matters as the reason ontologies can be shared and checked by machines rather than living only as a glossary in a slide deck.
A business ontology for sales, marketing and finance
A business ontology is the written agreement on what everyday business words mean, such as account, opportunity, customer and churned, across sales, marketing and finance. Without it, marketing may count trial sign-ups as customers, sales may count signed contracts and finance may count paying accounts, and every report disagrees.
This matters more with AI in the loop. An assistant asked how many customers churned last quarter will use whatever labels and records it finds; if the definitions differ between teams, the summary will be confident and wrong. Writing the definitions down, with one owner, is the low-tech version of an ontology and most of its value.
Where Pebbles Ai fits
Common questions
Is a CRM data model an ontology?
It is a light one: objects such as accounts, contacts and deals, with fields and links between them. It usually lacks the formal rules a full ontology states, such as what makes an account count as a customer.
Who should own a business ontology?
Usually revenue operations or a data team, with sign-off from sales, marketing and finance leaders. What matters is that one group can change a definition and everyone else can see the change.
What does ontology mean outside AI?
In philosophy, ontology is the study of what exists and how things can be grouped. AI borrowed the word for a written model of one domain's things and relations.
About this page. Written by the Pebbles Ai team for people who run sales, marketing and growth. Last updated .
